Re: Official Playstation 3 Blu-Ray player Thread
Quote:
But are these de-interlacing chips really built into all new TV sets? And do 1080i and -p machines really perform 3:2 pull down still? And would a TV set sense that an image is produced by a DVD player?
|
Sadly, most 1080p TVs and projectors on the market right now do *not* do proper frame-reconstruction when deinterlacing 1080i film-based content. They just "fill in the lines" using run-of-the-mill interpolation plotting (bobbing).
However, there are a handful of new 1080p sets and projectors that are incorporating real 3-2 pulldown reversal for full-quality 1080 deinterlacing for film content. This trend should continue as chip prices fall and awareness kicks in. We had the same thing with 480i deinterlacing... virtually no 480p TVs did 3-2 pulldown reversal back in the late 1990's when DVD first came on the scene. Then after a few short years it became almost standard. Faroudja's wise move to market their patended algorithm on an affordable chip that could be placed inside affordable consumer gear like TVs and progressive-scan DVD players was a large part of what made that happen. Hopefully we'll get some affordable chip-based 1080i deinterlacing solutions soon to help do the same thing.
Chances are that even your European gear has chips for 3-2 pulldown for 480 sources since much of the gear there is designed to be universal with NTSC/PAL dual-functionality.
for anyone with a 1080p display that does NOT do real deinterlacing for 1080i sources, but DOES allow 1080p input via HDMI, then those users would notice an appreciable improvement in clarity going with real 1080p input if they are getting a "real" 1080p signal from their BD or HD DVD player over HDMI (I qualify that bcs some 1080p hardware is just bobbing too!!!).
If a user has a display that does full deinterlacing for 1080i film, then they would probably see no difference going 1080i versus 1080p bcs their display is properly re-constituting each frame from the interlaced fields anyway.
BTW, your new DVDO processor most certainly does do full 3-2 pulldown reversal (sometimes called inverse telecine processing) for both 480 and 1080 interlaced sources mastered from 24 fps film. That's one of that processors claims to fame... in addition to being able to then output that 1080p signal in 1080p 24 fps form without adding back in the nasty 3-2 cadence judder for 60Hz. You picked the right product, I think it's the best-featured/highest-quality processor for the price.
Quote:
And would a TV set sense that an image is produced by a DVD player?
|
Yes... though the fact that they came from a "DVD player" isn't so much the key as is the answer to this question: "Is this 60Hz video something that started life as 24 fps film???"
Here's how it works (for both 480i as well as 1080i when implimented).
The TV set runs the video stream into a memory buffer that holds a series of fields (sometimes this causes a slight lip-sync audio delay as a result... another reason why the move to HDMI for audio as well as video). Back on track. The digital algorithm then looks at the series of fields in the memory buffer and tags any that are identical. In this way it can see if there's a 3-2 type pattern.
Here's what we mean by 3-2 pulldown/pattern: If we take 24 fps film and split each frame in half, we get 48 fields-per-second. So how do we expand that to fill a 60Hz signal for typical American TV sets? By repeating every 3rd field twice. I think it ends up something like this. If "A" are lines from the first frame and "B" are the lines from the second original frame:
Played back straight as a 48Hz signal:
A-odd, A-even, B-odd, B-even, C-odd, C-even, D-odd, D-even etc.
Ok... with 3-2 pulldown to make it into a 60Hz signal:
A-odd, A-even, B-odd, B-even, B-even*, C-odd, C-even, D-odd, D-even, D-even*...
I put a * next to the fields that got repeated twice in an effort to expand the 48 Hz signal to 60 Hz. This is field that the deinterlacing processor would see repeated... which would give it the clue it needs to recognize the 60Hz signal as film-source: If the 60 field-per-second video stream shows a 3-2 pattern of identical fields, then that's the "aha... this started life as 24 frame-per-second film!" answer the program was looking for. If so, then it drops out all the repeated fields and folds back together each actual "pair" of fields that form a real frame.
Now you've got a 24 fps progressive-scan signal.
From that point most processors repeat the frames back in the same 3-2 pattern to expand the 24 frames to 60 for a 60Hz signal. However, this step could be avoided and you could output the signal as 24p, 48p, 72p etc.
A similar process is used for PAL sources as well, but instead of looking for a 3-2 pattern the program looks for a 2-2 pattern.
If no "film" pattern can be found in the video stream, then most deinterlacers fall back on simple interpolation to fill in the missing lines. The better algorithms analyze the video to try to "merge" static areas of the picture while interpolating areas with motion to avoid combing (scan-line jaggies).
Quote:
| (BTW, my remark about many, slightly older TV sets not *really* accepting 1080p yet is certainly correct (well, "as far as I know" )). |
You're absolutely right on that count. Yet another "sigh, what were they thinking?!?" moment with cutting-edge gear that omitted a feature that seemed so obvious it's almost unbelievable that it wasn't provided. Sort of like the first generation of high-end digital projectors that lacked DVI/HDMI inputs! You should have heard the way the reps would try to spin that one: "VGA inputs are much higher quality than digital..."
p.s. here's some great links to help explain this whole bizarre thing about 3-2 pulldown with 60Hz video and 24 fps film:
http://www.zerocut.com/tech/pulldown.htmlhttp://en.wikipedia.org/wiki/Telecine
)).
